Reinforcement learning (RL) has demonstrated great potential for autonomous car-following control by directly optimizing key performance metrics, including safety, efficiency, and ride comfort. This study proposes an integrated RL-based car-following framework that combines trajectory prediction, reshaping of long-tail distributions, and uncertainty-aware decision-making. The trajectory prediction loss is redefined as a quantitative metric for scenario rarity and risk, facilitating a distribution-level data augmentation strategy that enriches safety-critical scenarios while preserving the naturalness of driving behavior. A probabilistic Gaussian LSTM predictor is further employed to model future vehicle motion uncertainty, which is explicitly incorporated into multi-step action evaluation through variance-weighted virtual rewards. Experiments on the highD dataset demonstrate that the proposed approach significantly improves robustness and safety, achieving substantial reductions in collision rates compared with baseline RL methods. Cross-dataset evaluation on NGSIM further confirms its generalization capability.

Timing performance and routability are two main goals of global routing. These two targets are mutually conflicting if we view and handle their effects independently. In this paper, we adopt a shadow price mechanism to incorporate the two issues into one unified objective function. We formulate global routing as a multicommodity flow problem. The objective function is the slack of congestion with the clock period as the delay limit from registers and inputs to registers and outputs. The multicommodity flow is expressed by a linear-programming formulation as a primal problem. We then convert the primal problem into a dual formulation using the shadow price as the variables. The shadow price of a net is the sum of its congestion price and timing price. The primal and dual formulation offers theoretical upper and lower bounds of the routing solution. Throughout the optimization process, the difference of the two bounds reduces, which provides the user's insight into the quality of the solutions. Based on the new formulation, this paper presents the UTACO algorithm for standard cell global routing.

Timing performance and routability are two main issues of global routing. In this paper, we adopt a shadow price mechanism to incorporate the two issues into one unified objective function. The shadow price of a net is the sum of its congestion price and timing price. Based on the new formulation, this paper presents the UTACO algorithm for standard cell (SC) global routing. The experimental results show that UTACO is efficient for both timing and congestion optimization.

We proposed a new global routing algorithm solving the net ordering problem. The algorithm uses random optimization methods to keep the equality of earlier routed nets and later routed nets in passing congested areas. It can find a solution independent of net ordering in short time. A global router is implemented in this method. Experiments show that the router performs much faster than Matula router while obtaining solutions with approximate quality.